the-decoder.com web signal

Google's Nano Banana 2.1 Halves Image Prices on Gemini 3.6 Flash

By Published Updated
4 sources tracking this story
Google AI Photo Generative AI ai-business

TL;DR

  • Google's official documentation confirms Nano Banana 2.1 removes function calling, structured outputs, caching, and Live API support present in prior versions, limiting which production workloads can migrate directly.
  • Runtime Wire establishes Oct 29 as only the earliest possible shutdown floor, not a final deadline, leaving developers no fixed date to plan a migration around.
  • Unite.AI's token consumption data (1,120 tokens per 1K input image) lets developers calculate actual per-call costs beyond the headline 50% output price reduction.

A 1K image on Google's new Nano Banana 2.1 costs $0.0336, down from $0.067 on Nano Banana 2. A 4K image drops to $0.0756, from $0.151. Nano Banana Pro, the quality-tier sibling, still sits at $0.134 per 1K image, per The Decoder's write-up.

The model is built on Gemini 3.6 Flash. It accepts up to 14 reference images in a single prompt and claims to keep up to four characters and ten objects consistent across conversation turns. It exposes minimal, medium, and high thinking levels, and renders at 1K, 2K, and 4K. In Google's framing, "version 2.1 improves visual quality, text rendering, character consistency across conversation turns, wide panoramas, and infographics."

Google's own benchmark numbers have 2.1 beating the pricier Pro variant on several metrics: 1106 vs 1011 on multi-character consistency, 1062 vs 990 on stylization, and 0.521 vs 0.265 on infographic accuracy. The source does not describe the rater pool or the benchmark suite behind those scores.

The caveat is Google's own. The write-up records that "in practice, Nano Banana Pro often still produces noticeably better images," particularly on color accuracy and proportions, with 2.1 positioned as "the more efficient counterpart" rather than a replacement.

Rollout spans the Gemini app, Google Search AI Mode, Google AI Studio, Flow, Stitch, Google Ads, and Google's Enterprise Platform. It lands in a dense week of Google releases we've been tracking, alongside the EmbeddingGemma 2 drop earlier this month.

What others are reporting

Coverage cluster as of 9h after publish

  1. Google AI Developer Docs Read →

    First-party spec sheet: the definitive capability list reveals missing features (no function calling, structured outputs, caching, Live API) absent from all press coverage.

    the latest high-efficiency image generation and conversational editing model
  2. Unite.AI Read →

    Adds token-level cost mechanics (1,120 tokens per 1K input image), safety red-team results, and known model limitations that reframe the announcement as an operational risk assessment.

    Nano Banana 2.1 as 'an update to Nano Banana 2 (Gemini 3.1 Flash Image),' built for high-efficiency image generation and conversational editing
  3. Runtime Wire Read →

    Developer-migration framing: Oct 29 is a floor not a deadline, and Google has not published the actual end date, leaving teams no fixed target to plan against.

    Teams using Gemini 3.1 Flash Image will need to check whether Nano Banana 2.1 works for their workloads before Google's listed earliest shutdown date.

Originally reported by the-decoder.com

Read the original article →

Original headline: Google Ships Nano Banana 2.1 on Gemini 3.6 Flash, Halves Image Prices and Lifts Reference Cap to 14